PyFAI v0.15.0

• 150 commits since last revision
• New tutorials on image inpainting, sensor thickness correction, ...
• Improve scripts
• Improve the new calibration GUI (pyFAI-calib2)
• Use scipy physical constants instead of hard-coded values
• Improved detector serialization and binning assessement
• Update documentation (images, better rendering of notebook & tutorials)
• Converge project with silx and fabio
• Remove generated rst- and C-files from repository
• This is the last version supporting python2.7
• Some feature as missing on debian7 due to old scipy

PyFAI version 0.14.2

Bugfix release with >110 commits:

• Fix seg-fault with manylinux1 wheels, in fastcrc module (thanks
  Thomas)
• Fix Qt4-Qt5 compatibility (thanks Vadim)
• Easier to understand geometry transformation (thanks Jon)
• Lower memory consumption, better cache management
• Unified debian packaging working for 6->9
• New detector: Mythen & CirPad (thanks Fred)
• Clean up debug code which avoid to use pyFAI-calib2
• pyFAI-calib2 now expect fabio >= 0.5
• Fix issue with metadata saving in 1d
• Fix performance regression with pyopencl >2015.2 (Thanks Andreas)
• pyFAI saxs and waxs scripts guess now the binning of the detector
  (thanks Fred).

PyFAI v0.14.1

New features:

• Fixes Debian 7 and 8 packages

PyFAI v0.14.0

New features:

• Graphical user interface for calibration (pyFAI-calib2)
• Goniometer calibration tools and multi-geometry enhancements
• Integration scripts and averaging scripts are now able to normalize the data
  from monitors found in the header.
• Propagate metadata information as part of the integrated data
• Common pre-processing factorization on Python, Cython and OpenCL
• Test clean up and acceleration (avoid tests on too large images)
• Many new tutorials http://pyfai.readthedocs.io/en/latest/usage/tutorial/index.html
• New averaging / integration methods:
  • Azimuthal median filtering
  • Azimuthal trimmed mean
  • sigma-clipping on azimuthal angle
  • Radial averaging
• Diffraction image inpainting to fill-up the gaps with plausible values.
• This release correspond to 572 commits
• Change of license: now all pyFAI is MIT license.

PyFAI v0.13.1

• Backport critical bug fix from 0.14-dev
  • Convert matplotlib mouse input into pixel coordinates (fix issue with numpy 0.12)
  • Fix eiger-mask script if h5py is not installed
  • Use "/usr/bin/env python" for scripts

PyFAI v0.13.0

• Global improvement of tests, packaging, code quality, documentation and project tools
• Scripts
  • Add support for multiframe formats on pyFAI-average
  • Add support for monitoring correction from header file (on pyFAI-average)
  • Add progressbar in the shell (on pyFAI-average and pyFAI-integrate)
  • Script drawMask_pymca is renamed into pyFAI-drawmask
  • Rework of the drawmask GUI using silx
  • pyFAI-drawmask do not have anymore hard dependency on PyMCA
  • pyFAI-integrate can now be used without qt dependency (--no-gui)
  • Fix the script to support both Python 2 and 3 (pyFAI-calib, pyFAI-benchmark)
  • Fix selection of units on diff-map (the user selection was not propagated)
• For users
  • More source code in MIT license
  • Update name and specification for cameras
  • Add cameras: Eiger500k, RaspberryPi5M, RaspberryPi8M
  • Fix Xpad S540 flat detector geometry
  • Fix definition of CeO2 calibrant
  • Add mask and flat on multi-geometry
  • Fix solid angle of the multi-geometry
  • Fix geometry processing for custom output space
  • Fix normalization factor and variance
  • Add support for Qt5
  • Add support for Debian 9 packaging
• For developers
  • Create common preprocessing for distortion correction
  • Create common image preprocessing using Cython (NaN filter, flatfield, dark, polarisation)
  • Refactoring of units module. It allows to register custom units.
  • Worker can now use Writer
  • Worker polarization argument is renamed into polarization_factor
  • Remove the dependency from python-fftw3, use numpy instead
  • Remove QtWebKit dependency
  • Fix un-correction of images using sparse matrix from scipy

PyFAI v0.12.0

• Continuous integration on linux, windows using Python 2.7 and 3.4+
• Drop support of Python 2.6, 3.2, 3.3 and debian6 packaging
• New radial output units: Reciprocal spacing squared and log(q)
• GPU accelerate version of ai.separate (Bragg & amorphous)
• Quantile filtering in pyFAI-average
• New graphical application for diffraction imaging
• Migrate to a common structure with silx (reorganize tests, benchmarks, ...)
• Extensions (binary sub-modules) have all been moved to ext directory
• Many improvements multigeometry integrators
• Compatibility with the copy module (copy.deepcopy) for azimuthal integrator
• Distortion correction works also for non-contiguous detectors
• Update documentation and provide advanced tutorials:
  • Introduction to pyFAI using the jupyter notebook
  • detector calibration
  • Correction of detector distortion, examples of pixel detectors.
  • calibrant calculation
  • error handling
• pyFAI-integrate can now be used with or without GUI
• Many new detectors (ADSC, Pilatus CdTe, Apex II, Pixium):
  • support for non-flat/curved detectors (Aarhus)
  • non-contiguous detectors (WOS Xpad)
• Include tests and benchmarking tools as part of the library
• Better testing.

PyFAI v0.11.0

• All calibrant from NIST are now available, + Nickel, Aluminum, ... with bibliographic references
• The Cell class helps defining new calibrants.
• OpenCL Bitonic sort (to be integrated into Bragg/Amorphous separation)
• Calib is available from the Python interface (procedural API), not only from the shell script.
• Many new options in calib for reset/assign/delete/validate/validate2/chiplot.
  • reset: set the detector, orthogonal, centered and at 10cm
  • assign: checks the assignment of groups of points to rings
  • delete: remove a group of peaks
  • validate: autocorrelation of images: error on the center
  • validate2: autocorrelation of patterns at 180° apart: error on the center function of chi
  • chiplot: assesses the quality of control points of one/multiple rings.
• Fix the regression of the initial guess in calib (Thanks Jon Wright)
• New peak picking algorithm named "watershed" and based on inverse watershed for ridge recognition
• start factorizing cython regridding engines (work ongoing)
• Add "--poni" option for pyFAI-calib (Thanks Vadim Dyakin)
• Improved "guess_binning", especially for Perkin Elmer flat panel detectors.
• Support for non planar detectors like Curved Imaging plate developped at Aarhus
• Support for Multi-geometry experiments (tested)
• Speed improvement for detector initialization
• Better isotropy in peak picking (add penalization term)
• Enhanced documentation on http://pyfai.readthedocs.org

many little features ...

• Convergence of pyFAI-calib and pyFAI-recalib, the later is now deprecated and will be discarded in the future
• Integration of an "empty" value for any bin without contribution. It is 0 by default.
• Python3 compliance (tested on 3.2 and 3.4) along with python 2.6 and 2.7
• The calibration can be launched from Python (not only from the shell script)
• Detectors are now able to guess the binning
• Support of multiframe EDF for diff_tomo
• Tested on windows 64 bits, MacOSX 10.6 and 10.6, debian 6, 7 and 8. Arm, PPC64LE, i386, amd64, with python 2.6, 2.7, 3.2 and 3.4. The test do not pass under windows 32bits due to the 2GB limit (nor under linux with 1GB of memory)
• Better support for detectors with gaps.
• Bunch of new calibrants (14 in total now)
• Sorting on GPU (works only under linux for now)
• Eiger detectors (masks, ...)
• Versoning tool (_version.py)
• Segmentation of the documentation: user/developer/administrator
• Clean up on the OpenCL side (factorization)
• Better workers and support for HDF5 input/output
• Tests are available in installed library: pyFAI.tests()
• Separation of third party code: argparse, six, ...
• Watershed based segmenter for peak-picking (work ongoing)

Improved peak-picking

Many small bug-fixes...